Up to 60% of people with bipolar disorder abuse drugs and alcohol, and, in contrast to the general population, bipolar patients are far more likely to meet criteria for substance dependence than abuse. Despite numerous studies documenting this frequent co- occurrence, it is unclear why substance dependence rates are so high in bipolar disorder. It is likely that more than one mechanism accounts for these elevated rates of substance dependence, which confounds studies of this problem. Therefore, one approach to clarify why substance dependence is so common in bipolar disorder is to define subgroups of patients with different mechanisms leading to this co-occurrence and then study how these subgroups differ. With this in mind, we propose that three patient groups defined by age at onset, course of illness, and life-events, will account for the excess of substance dependence in bipolar disorder: 1) patients in whom bipolar disorder initiates substance dependence, 2) patients in whom substance dependence initiates bipolar disorder; and 3) patients in whom both bipolar and substance use disorders develop as a result of a common risk factor (stressful life-events). The first major objective of this proposal, then, is to determine whether these patient groups can be identified in a sample of bipolar patients recruited at the time of their first manic episode, and to determine whether different substances of abuse are more commonly associated with each group. The second major objective is to determine whether these groups demonstrate different clinical outcome and familial rates of substance use and affective disorders. This study is novel in that it examines specific a priori-defined subgroups of bipolar patients with co-occurring substance dependence using prospective outcome, family history, and life events methodology to clarify why these two disorders so commonly co-occur. The study is further strengthened by a hypothesis driven design in a first-episode patient sample thereby eliminating the confounds of illness chronicity and prior treatment. The results of this study could have immediate impact on how treatment of this patient population is developed.